Multi-context gesture recognition
Gesturing is a natural part of our daily human interactions. Mobile, wearable, and embedded devices offer increasing opportunities to interact with computer systems through touch and mid-air gestures. During the last decade, a set of 2-dimensional stroke gesture recognizers, known as the “$-family”, took benefit of template-matching techniques to classify touch gestures. Despite years of research, many aspects of gesture recognition remain largely unexplored (e.g., 3-dimensional and mid-air gestures have been often excluded from the scope of $-family).
Pr. Jean Vanderdonckt and researcher Nathan Magrofuoco aim to achieve multi-context gesture recognition. To this effect, they study the influence of the context of use on gesture articulation and recognition (e.g. we articulate gestures differently from one context of use to another). Based on their observations, they plan to improve actual gesture recognition. Prominent collaborations allow them to identify, validate and react to their hypotheses: (1) with Pr. Radu-Daniel Vatavu (Université Ștefan cel Mare Suceava), main author of the most performant $-family recognizers, and (2) with Pr. Paolo Roselli (Università di Rome "Tor Vergata") whose researches aim to apply Clifford algebra to gesture recognition.